Voting with their Feet: Migration, Partisanship, and Party-Safe Elections in Florida

Transcription

1 Florida International University FIU Digital Commons FIU Electronic Theses and Dissertations University Graduate School Voting with their Feet: Migration, Partisanship, and Party-Safe Elections in Florida Rezwan Hussain Florida International University, DOI: /etd.FI Follow this and additional works at: Recommended Citation Hussain, Rezwan, "Voting with their Feet: Migration, Partisanship, and Party-Safe Elections in Florida" (2011). FIU Electronic Theses and Dissertations This work is brought to you for free and open access by the University Graduate School at FIU Digital Commons. It has been accepted for inclusion in FIU Electronic Theses and Dissertations by an authorized administrator of FIU Digital Commons. For more information, please contact

2 FLORIDA INTERNATIONAL UNIVERSITY Miami, Florida VOTING WITH THEIR FEET: MIGRATION, PARTISANSHIP, AND PARTY- SAFE ELECTIONS IN FLORIDA A dissertation submitted in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY in POLITICAL SCIENCE by Rezwan Hussain 2011

3 To: Dean Kenneth G. Furton College of Arts and Sciences This dissertation, written by Rezwan Hussain, and entitled Voting with Their Feet: Migration, Partisanship, and Party-Safe Elections in Florida, having been approved in respect to style and intellectual content, is referred to you for judgment. We have read this dissertation and recommend that it be approved Richard S. Olson Kevin A. Hill Dario Moreno Abraham Lavender Nicol C. Rae, Major Professor Date of Defense: November 7, 2011 The dissertation of Rezwan Hussain is approved Dean Kenneth G. Furton College of Arts and Sciences Dean Lakshmi N. Reddi University Graduate School Florida International University, 2011 ii

4 ABSTRACT OF THE DISSERTATION VOTING WITH THEIR FEET: MIGRATION, PARTISANSHIP, AND PARTY- SAFE ELECTIONS IN FLORIDA by Rezwan Hussain Florida International University, 2011 Miami, Florida Professor Nicol C. Rae, Major Professor Political scientists have long noted that Congressional elections are often uncompetitive, often extremely so. Many scholars argue that the cause lies in the partisan redistricting of Congressional districts, or gerrymandering. Other scholars emphasize polarization created by a fragmented news media, or the candidate choices made by a more ideological primary electorate. All these explanations identify the cause of party-safe elections in institutions of various kinds. This dissertation, by contrast, presents a structural explanation of uncompetitive elections. My theory is that population composition and patterns of migration are significant causes and predictors of election results in Florida. I test this theory empirically by comparing the predictions from four hypotheses against aggregate data, using the county as the unit of analysis. The first hypothesis is that Florida can be divided into clearly distinguishable, persistent partisan sections. This hypothesis is confirmed. The second hypothesis is that Florida voters have become increasingly partisan over time. This hypothesis is confirmed. The third hypothesis is that the degree of migration into a county predicts iii

5 how that county will vote. This hypothesis is partially confirmed, for the migration effect appears to have waned over time. The last hypothesis is that the degree of religiosity of a county population is a predictor of how that county will vote. This hypothesis is also supported by the results of statistical analysis. By identifying the structural causes of party-safe elections, this dissertation not only broadens our understanding of elections in Florida, but also sheds light on the current polarization in American politics. iv

6 TABLE OF CONTENTS CHAPTER PAGE 1 PARTISAN POLARIZATION AND PARTY-SAFE ELECTIONS REVIEW OF THE LITERATURE...8 1: PARTISAN POLARIZATION AND THE AMERICAN ELECTORATE...9 2: PARTISAN POLARIZATION AND ELECTORAL INSTITUTIONS...30 CONCLUSION A STRUCTURAL THEORY OF ELECTIONS : THE CONSTITUENCY BASIS OF POLITICS : AMERICAN VOTERS AS STABLE PARTISANS : COMPOSITION AND CONTEXT : MOVING AS POLITICAL BEHAVIOR : SECTIONALISM IN FLORIDA : FOUR HYPOTHESES ON FLORIDA ELECTIONS...78 CONCLUSION: A GENERAL THEORY OF ELECTIONS METHODOLOGY AND RESEARCH DESIGN : QUALITATIVE AND QUANTITATIVE METHODS : ANALYTIC METHOD : SOURCES AND TYPE OF DATA...99 CONCLUSION PARTISAN SECTIONS IN FORIDA INTRODUCTION : ELECTIONS IN FLORIDA: A COUNTY-LEVEL PERSPECTIVE : DO FLORIDA REGIONS FORM POLITICAL SECTIONS? : PARTISANSHIP AND FLORIDA ELECTIONS CONCLUSION EXPLAINING PARTISAN SECTIONS IN FLORIDA INTRODUCTION : EXPLAINING PARTISAN SECTIONS WITH STATISTICS : MODEL SPECIFICATIONS : REGRESSION RESULTS : RELIGIOSITY AND PARTISAN SECTIONS CONCLUSION BUILDING A GENERAL THEORY OF ELECTIONS : PARTISAN SECTIONS IN FLORIDA REVISITED v

7 2: EXTENDING THE RESEARCH : A GENERAL EXPLANATION OF ELECTIONS APPENDICES VITA vi

8 LIST OF TABLES TABLE PAGE 1: Methods in Scholarly Articles on American Politics, : Republican Margin of Victory, 2004 U.S. Presidential Election, by Florida County : Republican Margin of Victory, 2006 U.S. House Elections in which Both Parties Fielded a Candidate, by Florida County : Comparison of Republican Average Winning Margins, in U.S. Presidential Elections in the 1980s, 1990s, and 2000s, by Florida Region : Comparison of Republican Average Winning Margins, in Successive U.S. House Elections, in the 1990s and 2000s, by Florida region : Republican Average Margin of Victory in U.S. House Elections, Only Elections in which Both Parties Fielded a Candidate, by Florida Region : Effect of Uncontested U.S. House Elections on Republican Average Winning Margins, by Florida Region : Residence of County Population, Five Years Earlier, as a Percentage of the County Population 5 yrs of Age and Older, in : Control Variables Deployed in OLS Regression Models : Results of OLS Regression of Republican Average Winning Margins from U.S. Presidential Elections of 1988 and 1992, by 1990 U.S. Census Variables : Results of OLS Regression of Republican Average Winning Margins from U.S. Presidential Elections of 2000 and 2004, by 2000 U.S. Census Variables : Correlations (r) between Republican Average Winning Margins from U.S. House Elections of , and 1990 U.S. Census Variables : Results of OLS Regression of Republican Average Winning Margins from U.S. House Elections of , by 1990 U.S. Census Variables : Correlations (r) Between Republican Average Winning Margins from U.S. House Elections of , and 2000 U.S. Census Variables : Results of OLS Regression of Republican Average Winning Margin from U.S. House Elections of , by 2000 U.S. Census Variables vii

9 16: Correlations (r) between Democratic Winning Margins from 2004 U.S. Presidential Election, and 2006 American Community Survey Variables : Results of Simultaneous OLS Regression of Democratic Winning Margins from 2004 U.S. President Elections, by 2006 American Community Survey Variables : Results of Hierarchical Regression of Democratic Winning Margins from 2004 U.S. Presidential Election, by 2006 American Community Survey Variables : Results of Hierarchical Regression of Democratic Winning Margins from 2004 U.S. Presidential Election, by 2006 American Community Survey Variables viii

10 LIST OF FIGURES FIGURE PAGE 1: Winning Margins of the Two-Party Vote in 2004 U.S. Presidential Election, by Florida County : Florida County Partisanship, using Average Winning Margin of the Two-Party Vote in U.S. Presidential Elections of 2000 and : Winning Margins of the Two-Party Vote in 2006 U.S. House Elections, by Florida County : Comparison of Average Winning Margins of the Two-Party Vote in U.S. Presidential Elections in the 1980s, 1990s, and 2000s, by Florida Region : Average Winning Margins of the Two-Party Vote in U.S. Presidential Elections of , by Florida Region : Comparison of Average Winning Margins of the Two-Party Vote in U.S. House Elections in the 1990s and 2000s, by Florida Region : Comparison of Average Winning Margins in U.S. House Elections in the 1990s and 2000s Only Elections Contested by Both Parties, by Florida Region : Graphs of Winning Margins in Elections for State and U.S. Federal Government Offices, Between 1980 and 2006, for Five Large Florida Counties : Graphs of Winning Margins in Elections for U.S. President, U.S. House, and Florida Agriculture Commissioner, Between 1980 and 2006, by Florida Region : Florida s Partisan Sections, by Average Winning Margins from U.S. House Elections of : Florida s Partisan Sections, by Average Winning Margins from U.S. House Elections of ix

11 CHAPTER 1 PARTISAN POLARIZATION AND PARTY-SAFE ELECTIONS 1

12 Over the last two decades, observers of American politics have noted that American politics has become more polarized, and that members of the U.S. Congress appear to engage in much more partisan bickering. Scholars have found that members of the two Congressional parties are more inclined to vote along party lines, and more inclined to vote along an ideological divide. Republican members of Congress have become more cohesive and conservative in their voting, while Democratic members have become more cohesive and liberal in their voting. This increasing partisan polarization has become a cause for concern to many Americans. Many academics and members of the general American public worry that polarized politics will make America harder to govern, or even render American government ineffectual, as the affairs of state grind to a halt amidst partisan gridlock, and necessary legislation is passed only after partisan brinkmanship. Not surprisingly, scholars of American politics, as well as the media, have devoted considerable time and energy into describing, explaining, and theorizing about the causes and consequences of this polarization in American politics. To explain this development, scholars have looked more closely at the institutions and processes of American elections. This focus is eminently sensible, because elections are how members of Congress are selected. Since elections involve voters, one possible cause of polarization in Congress is the attitudes and behavior of the American electorate. Have American voters become more polarized? This question 2

13 has fomented a lively debate, as scholars argue over whether a culture war is underway, driven by emotive issues such as abortion or gay marriage. Other scholars have looked to elites and institutions as likely causes of partisan polarization. These scholars are interested in investigating whether those who run for office, who emerge as candidates in American elections, have become more polarized always assuming, of course, that previously moderate candidates do not suddenly change their stripes and become polarized ideologically after being sworn into office. Accordingly, some scholars have looked into the candidate selection process as the cause of polarization, and in particular, into primary elections. These elections, they argue, are dominated by ideologically polarized party activists, who select more polarizing candidates to run in the general election. Other scholars have examined the structural features of the general election itself, and in particular the uncompetitive nature of many of these Congressional elections. This focus on elections also makes sense, since if elections are uncompetitive, there is little reason for an incumbent to spend time or effort appealing to voters who did not vote for him or her; the incumbent can safely ignore these voters, and concentrate on tending to those the voters who have already voted for him or her. According to this line of thinking, uncompetitive elections contribute to polarization because incumbents generally ignore the median, or moderate, voter, and concentrate on appealing to the more polarized party base. 3

14 Originally, this incumbency advantage was thought to derive from advantages inherent in the office of U.S. Representative, such as the resources and visibility afforded to the incumbent, These strategic incumbents, scholars argued, learned how to please their constituents with the votes they cast in Congress, and their style of communicating at home in the districts, and deterred high quality challengers by raising large amounts of campaign funds an activity aided by their incumbency status. The result was to make elections safe for incumbents, and, by extension, the incumbent s party. In recent years, however, scholars have begun to rethink incumbency advantage. They noticed, for instance, that when an incumbent retired, the seat almost always remained in the hands of the party of the departing incumbent. So did the incumbent hold that seat because of the advantages of incumbency, or because of the advantages of being identified with the right party? In other words, was the incumbency advantage personal, or partisan? It was a puzzle, because challenges are rarely successful, and the number of competitive districts continued to decline, into the first decade of the twenty-first century. If Congressional elections are uncompetitive because one or other party enjoys the support of a clear majority in the district, election after election, what could account for this partisan incumbency advantage? Many scholars identified the cause in redistricting: the practice of redesigning districts after the decennial Census to account for population changes and make electoral districts more equitable in 4

15 population size, pursuant to Federal law and Supreme Court ruling. The process of redesigning districts, these scholars argued, was being conducted for partisan advantage: the partisan political elites who controlled the process were gerrymandering districts in order to pack those likely to vote for the opposite party into a limited number of districts, leaving more districts that are safe for the candidates of their own party. The redistricting explanation is enormously popular, among both scholars and media outlets, often appearing in the media and popular discourse whenever the topic turns to Congressional elections. Unfortunately, as often happens with such nostrums, the partisan redistricting explanation of party-safe Congressional elections is not clearcut: elections in districts that have not been redesigned display many of the same trends as districts that have, and the U.S. Senate has become almost as polarized as the U.S. House of Representatives, and many Senate elections have become just as party-safe. But even if redistricting is not the reason why many Congressional districts display a persistent partisan advantage, the way voters are distributed may still confer an electoral advantage to a specific party. In other words, it may not be because district boundary lines are being moved by legislators, commissions and judges, but because voters are moving themselves, from one place to another. Are American voters, who live in a famously mobile society, making residential choices that are creating electoral districts with distinct and persistent partisan profiles? 5

16 In this dissertation, I will construct and test the prime hypothesis that the distribution of voters is the cause of party-safe elections, using Florida as a test case. I will examine whether the American political landscape displays distinct partisan sections geographical areas where one party has an embedded and persistent advantage and whether this sectionalism can be linked to Congressional election outcomes, and party-safe elections in particular. My argument is that to understand American Congressional elections, we must understand American voters, and the way they are distributed across the political landscape, within the nation and within individual states. My argument identifies the behavior and attributes of the American voter as the cause of party-safe elections. My objective is not to supplant existing explanations of party-safe elections, explanations that emphasize elites and institutions as the causal variables, but to complement them. If we can better understand the causes of uncompetitive elections, we gain a better understand of increasing partisan polarization, which, I believe, is a consequence of party-safe elections. As such, this study not only contributes to the scholarly literature on Congressional elections and American electoral behavior, but also contributes to our real world understanding of American politics. In the next chapter, I will review the scholarly literature on district competitiveness and polarization in elections for the U.S. House, and elaborate on the some of the points made in this introduction. In chapter 3, I will specify my explanation of partysafe elections, one that takes the form of an empirical theory. In chapter 4, I discuss 6

17 various elements of my research design and method, such as my choice of units of analysis, and the observations that constitute my data sets. In chapters 5 and 6, I discuss my findings: in chapter 5 I describe the partisan sections that emerge in Florida, and in chapter 6 I build explanatory models of these partisan sections. I conclude with an overview of what I have done in this study, and describe one way to develop and extend it. 7

18 CHAPTER 2: REVIEW OF THE LITERATURE 8

19 In this chapter, I will review what scholars have said about partisan polarization and electoral competitiveness in American politics. Both polarization and electoral competitiveness are multidimensional phenomena, involving institutions, elite behavior, and structural factors such as public opinion. For analytic and organizational purposes, I will first discuss voter behavior, which is central to public opinion literature, and then I will discuss the literature on elites and institutions. Regardless of the order of discussion, the central question remains the link between elites and the mass public. The question of representation has been a central concern in the study of American politics; I do not expect to resolve any major controversy here. My objective is to review the relevant literature that can help us understand America s polarized politics. I begin, in the next section, with a brief synthesis of what political scientists have said about American electoral behavior over the last three or four decades. I will paint a picture with very broad strokes. My goal in Section I is to introduce the topic of American elections and American political parties, and contextualize my research question. 1 PARTISAN POLARIZATION AND THE AMERICAN ELECTORATE The scholarly explanation of elections and partisan polarization encompasses various processes: a realignment of the balance of electoral support for the two major parties 9

20 in the American political system, the behavior of elites as cause and effect of these changes, institutional factors such as Congressional reforms, primary elections, and the redistricting revolution. In short, the panoply of variables political scientists often summarize as structure and agency. 1 Scholars often describe the landscape of contemporary American electoral politics as the outcome of a process that began with the defection of conservative white Southerners from the Democratic Party to the Republican Party over differences regarding the Civil Rights agenda. They go on to argue that this shift in electoral politics, typically referred to as the Southern realignment, made the two parties more ideologically coherent entities, more clearly differentiated from one another along ideological and policy profiles, as opposed to geography (Carmines and Stimson 1989; Black and Black 2002). According to the literature, this realignment was a two-stage process. First, Southern whites began voting Republican, initially in presidential elections (in the 1960s), and then in Congressional elections (in the 1980s), transforming ( realigning ) the American South from a Democratic into a Republican Party stronghold by the 1990s. The Southern realignment was reinforced by the migration of many middle-class whites from the Northeast and Midwest to the Sunbelt (Black and Black 1987), a process partly motivated by technology and infrastructure, such as air conditioning and the interstate highway system (Polsby 2004). The realignment 1 This phrase, popular with many scholars, conflates structure with institutions. Since this distinction is central to my analysis, I will not employ this phrase again. 10

21 not only re-shaped the nation s political geography, but also led to the rise of a new generation of national political leaders from the South, such as Newt Gingrich (Kruse 2006). This realignment of the electorate led to changes in the two parties in Congress. Over several election cycles, conservative white Southern Democrats became less important in the House and Senate, and were increasingly replaced by Republican and Democratic African American members. The new Democrat cohorts replaced a personalistic, committee-based chamber dominated by seniority and a bipartisan conservative coalition into a more partisan, majoritarian chamber dominated by party leaders and whips (Rohde 1991, Polsby 2004). The transformation of Congress did not happen overnight: it took two waves of reforms, in the 1970s, under Democratic leadership, and in the 1990s, under Republican leadership (Sinclair 1995; Rae 1998). Rank-and-file members, for their part, gave Congressional party leaders the power to make these changes because they trusted the latter to serve party ideology, rather than sectional or constituency interests (Sinclair 1995). The outcome was a more polarized, partisan institution, characterized by greater party-line roll-call voting (Poole and Rosenthal 1997), gridlock (Brady and Volden 2006), polarizing leaders (Sinclair 1995) and increased partisan rancor in general (Jacobson 2004). The partisan polarization within Congress cued ideological polarization within the American electorate (Abramowitz and Saunders 1998; Jacobson 2000a; Black and Black 2007). Increasing ideological coherence within the Congressional parties 11

22 with liberals aligned with the Democratic Party, and conservatives aligned with the Republican Party, taught voters that party labels meant something that they indicated and predicted how representatives would vote or act once in office (Jacobson 2000a), and encouraged voters to align their party identification with their preferences on various issues (Abramowitz and Saunders 1998; Carmines and Layman 1997). Whether described as ideological realignment (Abramowitz and Saunders 1998), sorting (Fiorina and Levendusky 2006), or just polarization, the result was greater ideological coherence and party loyalty within the American electorate (Jacobson 2000b), which became increasingly divided politically, along regional, racial, religious, ethnic, income, gender and cultural lines (Carmines and Stimson 1989, Layman 2001, Brewer and Stonecash 2007, Black and Black 2007, Bartels 2008). Scholars found evidence of this increasing division in various domains, and among various groups. There are three such groups that, for many political scientists, matter most for electoral politics: members of Congress, party activists, and voters. The most compelling evidence of polarization is probably in the U.S. Congress. The Republican and Democratic parties in Congress appear more solidly opposed to one another, more cohesive, and more ideologically distinct from each other. Measures such as party votes, where a majority of Republicans oppose a majority of Democrats, and party unity scores, which measure how often a member of Congress votes with his or her party, indicate that Republicans and Democrats in Congress began voting 12

23 more along party lines in the 1980s (Rohde 1991). Other measures, such as interest group scores of individual members (Fleisher and Bond 2000), or the DW-Nominate scores calculated by Poole and Rosenthal, which capture the ideological distance between members using all roll call votes (Poole and Rosenthal 1984, 1997), indicate that members of Congress have become more ideologically polarized. In short, both Republican and Democratic members more become partisan and more ideological; the two parties have become more cohesive. The polarization appears to extend to party activists. Some scholars argue that ideological differentiation, or the distance between Republicans and Democratic activists on economic and social issues, has increased (Aldrich 1995; Layman 2001; Fiorina 2002). Such polarized activists contribute to party polarization, according to these scholars, by virtue of the influence they exert over the candidate selection process, one that throws up more polarized candidates for the general election, and sends more polarized office-holders to Congress (Burden 2001; Fiorina et al. 2006; Brady and Han 2007; Masket 2009). And finally, polarization appears to extend to the electorate. Scholars first noticed increasing partisanship in presidential voting in the 1980s (Miller 1991; Bartels 1992), and subsequently in Congressional voting (Bartels 2000). Measures of public attitudes toward the parties, such as the feeling thermometer in NES surveys, also indicate that voters became more partisan beginning in the 1980s (Hetherington 2001), and the proportion of voters who self-identified as independents a measure 13

24 that some consider to be the most revealing measure of aggregate partisanship - fell (Bartels 2000). At the same time, ticket splitting, or voting for different parties for president and Congress, declined (Hetherington 2001; Stonecash 2006). The increasingly partisan behavior, according to some scholars, is evidence of a polarized electorate, as voters increasingly align their partisanship with their preferences on various social and political issues (Jacobson 2000b). Political scientists, however, are as contentious as any group of scholars, and many of the observations offered in this capsule summary have been simply grist to the mill for scholarly argumentation and debate. Scholars argue over whether the electorate is actually polarized, or whether only elites are polarized, over what causes polarization, and, as always, over what links elites to the mass public the central question in the public opinion literature for decades (Converse 1964; Delli Carpini and Keeter 1996). These issues dominate the scholarly literature on American public opinion and the American party system, and the remainder of this chapter is devoted to these debates. To organize the discussion conceptually, I will conceptualize the electoral arena using three constituent elements: voters, the elites who seek to influence or earn those votes, and the electoral institutions through which elites act. I have presented polarization as an outcome of electoral politics; consequently, I will conceive of polarization as a product of the interaction and mutual reinforcement between voters, elites, and institutions, which refer to both elections and political parties (Jacobson 2000a). 14

25 I begin by discussing the electorate. The logical way to do so is by reviewing the theoretical framework political scientists have used for decades when thinking about American elections, political parties and even political change in general: realignment theory. Then, I will focus on the link between elites and voting behavior. I will end the section with a discussion of the so-called culture war, which can be thought of simply the latest version or iteration of the complex and never-ending interaction between elites and voters. 2.1 Realignment and Partisan Polarization In the original formulation of this theory, realignment occurs when stable patterns of partisan alignment and electoral coalitions are sundered by a powerful, polarizing issue, one that displaces the existing conflict, cuts across pre-existing political cleavages, and realigns the electoral coalitions of the two parties (Key 1955). Key first proposed that realignments occurred convulsively, in a critical election, such as those of 1860, 1896, or 1932 (Key 1955), but later expanded the concept to include longer secular realignments that take place over several election cycles (Key 1959). The literature has grown enormously since then, but realignment theory still refers to one of two processes: the rise of a new issue around which political coalitions divide, or the rearrangement of partisanship within the electorate (Stonecash, Brewer and Mariani 2003). 15

26 As with many ubiquitous concepts, however, scholars are divided over different aspects of the concept, such as how long realignment takes, whether the cause is conversion or mobilization, whether the cleavages are societal or issue-based, whether cleavages displace each other, or accumulate, and whether realignments are national or sectional phenomena (Petrocik 1981, Carmines and Stimson 1989; Dalton et al. 1984; Bensel 1984). For the purpose of my research question, in which the link between voters and elites is a central concern, three popular accounts of how and why realignment occurs are worth noting. Issue Evolution In classical realignment theory, as Sundquist (1983) remind us, party elites are the first to respond when a polarizing issue emerges; their actions motivate voters to do one of two things: review their own stance on the issue, and change their party alignment (conversion), or, if they are independents, to choose a party (mobilization). Carmines and Stimson argue, however, that the Southern realignment was not a function of conversion or mobilization, but of generational replacement - the entry of younger voters into the electorate with different attitudes (Carmines and Stimson 1989). They suggest that this issue evolution began in the South in 1964, as disagreements regarding Civil Rights and racial integration pushed Southern whites out of the Democratic and into the Republican Party, simultaneously motivating newly empowered black voters to identify with the Democratic Party. 16

27 Ideological Realignment According to Abramowitz (1994), this issue evolution explanation confuses correlation with causation. He argues that racial attitudes are really a correlate of ideology the broad set of attitudes toward the proper role of the state in providing welfare and national security. Ideology drives partisanship and partisan realignment, not race. These two issues first became salient in the 1964 presidential election, and have driven partisanship ever since (ibid.). As in classical realignment, Abramowitz acknowledges that party leaders act as catalysts, but he attributes more agency to voters, who contribute to polarization by realign themselves in two ways: first, by matching their self-identification as liberals or conservatives (their ideology) with preferences over a range of issues, and then by matching that ideology with a specific party (Abramowitz and Saunders 1998, 2008). Conflict Extension In a more recent approach, Layman and Carsey (2002) take issue with ideological realignment for exaggerating the ability of an electorate that is essentially ideologically incoherent to mimic the ideological coherence of elites. They also take issue with the central classical realignment concept of conflict displacement, that realignment occurs when a new, polarizing issue displaces an older issue. Conflict displacement, the authors argue, may describe how voters respond to new issues, but elites do not displace conflict, but extend it they polarize on several issues simultaneously. The consequence of this conflict extension is to heighten political 17

28 polarization, as one polarizing issue is piled on top of another, reinforcing the previous cleavage, rather than displacing it or cutting across it (ibid.). The realignment literature thus offers different interpretations on the link between voters and elites. The debate continues because, whereas most political scientists accept that political office-holders and activists appear to possesses a set of coherent beliefs about politics, and can be placed along a liberal-conservative ideological continuum (Poole and Rosenthal 1984, 1997), the voting public is an entirely different matter. Ever since Converse (1964) first reported that American voters were generally ill-informed about political issues, and, with the exception of selected groups ( issue publics ), who took an interest in specific issues but did not possess any coherent set of political beliefs, scholars have been bedeviled by the question of whether or not voters possess any coherent understanding or conception of politics, or ideology (Lewis Beck et al. 2008). 2.2 Is the Electorate Polarized? Those who argue that voters have become more polarized take one of two approaches, which I will refer to as participatory and informational. The first approach builds on the concept of political participation, a very broad term that encompasses not just voting but also a host of behaviors that typically attend elections. These behaviors can be cognitive, such as being aware of candidates and 18

29 issues, and being able to distinguish between parties, affective, such as taking an interest in the election outcome, or actual political activity, where activity is typically defined as engaging in at least one act beyond voting, such as volunteering, or helping to raise campaign funds. Abramowitz points to greater levels of these types of political participation to support his claim that the contemporary electorate is more ideological than the one Converse encountered in the 1960s (Abramowitz 2006). Looking at the National Election Studies (NES) data for the 2004 general election, he concluded that levels of politically activity, awareness of differences between the parties, and interest in the election outcome were all at record levels (Abramowitz and Saunders 2008). Political participation, in this approach, takes the form of a proxy measure for ideology. The most popular approach taken by those arguing that the electorate is more polarized is one that I have labeled informational. If the electorate is more polarized today, these scholars maintain, it is because voters are taking their cue for how to think or behave politically from signals they receive from political elites - if those signals are sufficiently clear and strong (Brody 1991; Page and Shapiro 1992; Zaller 1992). Scholars who make this argument usually add that these signals are typically reinforced by media references to polarized politics (since voters typically obtain information from politics from the mass media); the result is greater ideological awareness and coherence within the electorate. This cue-taking hypothesis is really the theoretical core of the various realignment theories just discussed - even Abramowitz s conception of ideological realignment. 19

30 Unfortunately, it is not at all clear that voters do indeed respond to elite cues, for the simple reason that they do not pay enough attention to politics (Converse 1964; Delli Carpini and Keeter 1996). The exception is the minority of voters who are politically informed, the so-called strong party identifiers (Zaller 1992). According to Layman and Carsey (2002), these strong party identifiers respond to elite cues on an issue in one of two ways, depending on how much the issue matters to them, or the saliency. If the issue is important, these partisans may defect to another party. This process is consistent with classical realignment theory, where cross-cutting cleavages motivate voters to cross party lines; this response creates a ceiling, or cap, on partisan polarization (Sundquist 1983). But if the issue is not salient, partisans will modify their preference on that issue to better align it with their party s platform. This process extends political conflict, rather than displacing it; it encourages party elites to move even farther apart, and heightens partisan polarization (Layman and Carsey 2002). Do Voters take Cues? The elite cuing idea appears plausible. Nevertheless, it is not entirely consistent with some empirical observations about mass political behavior. For instance, repeated surveys conducted every decade since the 1970s in the World Values Survey project, administered by the University of Michigan, describe a mass public in the U.S. that is increasingly individualistic, non-hierarchical and elite-challenging in their political style and behavior (Inglehart 1997, 2005). According to Inglehart, this expressive politics is manifest at the individual-level in petition signing, recall and 20

31 ballot initiatives, and the increasing emphasis on post-material issues, such as the environment (ibid.). The expressive politics hypothesis dovetailed neatly with inquiry into partisan de-alignment, which dominated the scholarly agenda during the 1970s through the 1990s, but it does not fit so neatly with the hypothesis that the public is following the lead of elites. So do Americans favor elite-challenging expressive politics, or are they taking their cues from polarized partisan political elites, such as activists and party leaders? Logically, it is one or the other. In short, the World Values Survey findings are at odds with the idea of an electorate increasingly polarized by elites. The elite-cueing explanation is also problematic if we consider the behavior of elites the party activists, the candidates for elected office, and the professional political class hired to help them do so. If the electorate follows cues from these elites, why do contemporary political campaigns expend vast resources gathering and analyzing data on every conceivable aspect of voters lives? Why do candidates spend everincreasing amounts on micro-targeting potential voters (Baker 2008)? The critical importance of demographic research and grass-roots organizations in contemporary elections, not to mention the obsessive attention to opinion polling, is not consistent with the elite-cueing hypothesis. These practices imply that it is elites who are looking to the mass public for cues. These features of the voter-based, or public opinion, perspective on polarization is amply illustrated in the debate that has raged over what many consider the dominant 21

32 cleavage in contemporary American politics: the increased salience of cultural issues, and the so-called culture war. 2 Is this an example of political conflict being extended, rather than displaced? It is worth taking a closer look at this debate, for it illustrates the strengths and weaknesses of the argument that contemporary polarized politics stems from a polarized American electorate. 2.3 The Culture War Hypothesis There are two elements to the culture war hypothesis. The first is that there is a religious divide in the American public. Unlike the religious cleavage of a century ago, defined by the Protestant Catholic divide, (Sundquist 1983; Brewer and Stonecash 2006), the contemporary religious divide is not between denominations, but within them. It is one that pits religious traditionalists, or orthodox in Hunter s terminology, against modernists, or progressives (Hunter 1991). The former, associated with evangelical Christians, consider the Bible to be authoritative; the latter, associated with mainline churches, consider the Bible to be a social artifact, and consider political institutions, embedded in distinct cultural contexts, to be authoritative (Wuthnow 1988; Hunter 1991; Layman and Carmines 1997; Layman 2001). 2 The sociologist John Davison Hunter is credited with coining the term (1991); Pat Buchanan introduced the term to mainstream America in his speech at the 1992 Republican National Convention. 22

33 The second element to the culture war hypothesis is that this religious conflict has spilled over from the religious realm to the political realm, and religious beliefs now influence not only personal issues (family, gender and sexuality) but also a range of attitudes regarding the proper role of government in society, or ideology (Wuthnow 1988; Hunter 1991). This spillover into politics, which Hunter (1991) refers to as isomorphism, has two consequences: (i) it creates an ideological electorate and (ii) it is creates particularly intractable political attitudes, since religious beliefs tend to be incommensurable. The result is that, instead of negotiation and compromise over differences, political discourse becomes an unyielding struggle, or war (ibid.). Layman and Carmines (1997) examined NES data, and, consistent with this perspective, concluded that traditional religious values were the strongest predictor of the presidential vote - a measure often taken as a proxy for ideology. Researchers analyzing the World Values Survey have concluded that church attendance is the strongest predictor of ideology, with greater attendance being positively correlated with politically conservative attitudes - not only in the U.S., but also throughout the industrialized world (Norris and Inglehart 2004). As with the accounts of polarization in general, political activists are identified as the prime catalysts of this culture war. In one influential work in this school of thought, Layman (2001) argued that it was activists and party leaders who turned the religious divide into one that extended into partisan politics. In this argument, conservative activists felt threatened by the changes promoted by post-materialists and other 23

34 secular groups in contemporary society, and pressed party leaders to exploit the great divide between religious traditionalists and secularists in order to mobilize conservative voters. Party leaders obliged, and voters, taking their cues from family, friends, and community, re-aligned their support for the two parties: traditionalists aligned with the Republican Party, and more secular voters with the Democratic Party (ibid.). 2.4 Is There a Culture War? How does this culture war hypothesis stack up against empirical evidence? DiMaggio and his collaborators scoured two decades worth of survey responses in the General Social Survey (GSS) and NES data and found that, contrary to the culture war predictions, the public s opinions on social issues were converging, rather than polarizing (DiMaggio, Evans and Bryson 1996; Evans, Bryson and DiMaggio 2001). The one exception appeared to be abortion - although Mouw and Sobel (2001) dismissed even that seeming evidence of polarization as basically measurement error. Davis and Robinson (1996) for their part, found that religiosity correlated with family and gender-related issue attitudes, but not political attitudes. Instead, as predicted by classical realignment theory, they found cross-cutting cleavages - attitudes that cut across party lines; they found that religious traditionalists were just as likely to be economically conservative as they were to be economically liberal. The authors do find that religious elites were more doctrinaire than rank and file congregations, just 24

35 as political elites were more ideological than the mass public, but that, on the whole, religious conservatives in the U.S. resemble the electorate at large: divided along race, sex, class and age, with no consistent issue attitudes (Davis and Robinson 1996). Partly inspired by these findings, Fiorina published arguably the most influential and widely-read critique of the culture war thesis in According to Fiorina, the distribution of opinion within the electorate (polarization) has been unchanged for decades: the relative proportion of liberals, conservative, or moderates within the electorate has remained steady since the 1950s (Fiorina et al. 2005). And moreover, the American public remains much as Converse depicted it in 1964 ill-informed about political issues in general, divided over specific issues, and lacking in any coherent ideology - religious or otherwise (Fiorina et al. 2006; Fiorina and Levendusky 2006). What has changed, according to Fiorina, is the behavior of party leaders, candidates and activists, behavior that can be explained by the process by which such elites are selected: primary elections dominated by ideological party activists (Fiorina 2002, Fiorina et al. 2006). Since voters are restricted to the choice of elites who emerge from these primaries structure the vote choice, they end up having to choose between party candidates with polarized positions on issues that primary activists hold dear, such as abortion, gay marriage, or gun control. In short, the electorate is sorted along partisan lines, and this sorting creates the misleading impression that the electorate is polarized (Fiorina et al. 2006, Levendusky 2009). 25

36 While scholars such as Abramowitz and Fiorina focus on indicators such as political participation or the distribution of attitudes, and seek to explain political behavior in terms of political orientations, a burgeoning literature in political science investigates whether political behavior can be explained by psychological attributes, such as personality. One recent study examined links between parenting values or family structure and partisanship or vote choice. Using a parenting scale developed from a series of questions in the NES about childrearing, Hetherington and Weiler (2005) test the hypothesis that the American electorate can be sorted by a strict parent/nurturing parent binary. They find a clear correlation between strict fathers and Republican partisanship, and between nurturing parents and Democratic partisanship (Hetherington and Weiler 2005). I will revisit this interesting new line of research into the electorate in the concluding chapter. The culture war controversy notwithstanding, most scholars acknowledge that religion has long been a source of political cleavage in American politics (Sundquist 1983). The question is whether, after being quiescent for a period of time, religion has resurged as a factor in American politics, and if so, how it contributes to partisan polarization. But why did the religious divide subside in the first place? 26

37 2.5 Income and Partisan Polarization According to some scholars, the religious divide was displaced in the twentieth century by a new, overriding cleavage in American politics: economic class (Glaeser and Bryce 2006). Economic class as the dominant cleavage in American politics is typically attributed to the Great Depression and the ensuing New Deal realignment of American politics, in which the working and lower middle-classes aligned with the Democratic Party of FDR, and the upper middle and upper classes aligned with the Republican Party (ibid.). This New Deal alignment is familiar to most Americans in the commonplace stereotype of Republicans as the party of the rich, and Democrats as the party of the little guy. I will not belabor this point, partly because, unlike the difficulties associated with measuring ideology, which relies on subjective responses to survey questions, not to mention the continuing debate about whether ideology actually matters to American voters, there is less controversy about whether class is a factor in electoral politics, if only because class can be measured using more objective criteria. In fact, there is a range of criteria for scholars to choose from since class is a multidimensional concept, and can be defined using income, education, or occupation as indicators. Nevertheless, much of the literature has traditionally defined class in terms of income. Bartels, for instance, finds that whereas class defined by college education has become less relevant as a factor in voting over the past half-century, class defined by income has become more relevant (Bartels 2008). 27

38 Class-oriented scholars argue that partisan polarization stems not from attitudes toward race or culture, but from differences in (income-based) class (McCarty, Poole and Rosenthal 2006; Bartels 2006, Gelman 2008). McCarty et al., for instance, argue that racial attitudes in the South cannot explain polarization in Congress, since the demise of liberal Republican members outside the South is clearly observable beginning in the 1970s (ibid.). The Southern realignment, in their view, was simply an extrapolation of an income-based polarization that was already underway nationally. According to their analysis, polarization in the South actually trailed the rest of the nation (ibid.). To explain the observation that voting in the U.S. House of Representatives since the 1970s clearly maps onto a liberal-conservative dimension, they identify personal income within a district as the cause: the wealthier the district, the more the Representative in Congress hews to the Republican Party line (McCarty et al. 2006). Scholars who emphasize class rather than ideology argue that the influence of economic class, or income, on election outcomes has been increasing in recent decades, and that the correlation between class and voting behavior is now considerably higher than in the 1950s (Brewer and Stonecash 2006; Bartels 2006). These scholars point to polling evidence that income is a strong predictor of Republican partisanship, and that the New Deal alignment is very much intact: poorer whites tend to vote Democrat, whereas middle and upper income voters have become ever more Republican (Bartels 2006, Gelman 2008, McCarty et al. 2006). To the extent that white voters without college degrees have become more Republican over 28